Browse Articles

The Indian Ocean disaster is a climate tragedy — and needs more attention

Waveguide design for a TFLN platform at 1064 nm for applications in spacecom and spectroscopy

PyUAT: An open-source Python framework for uncertainty-aware, efficient, and scalable model-driven cell tracking

Tracking individual cells in live-cell imaging provides fundamental insights into phenotypic heterogeneity and cellular responses to environmental change. However, microbial cell tracking is particularly challenging, as cell growth is characterized by stochastic cell movements and frequent divisions, while time-lapses are recorded at limited frame rates to avoid counterfactual results. Here, we investigate how probabilistic Uncertainty-Aware Tracking (UAT), a paradigm based on statistical models of cell behavior, robustifies tracking quality under such challenging conditions. Using , the first open-source implementation of UAT, we systematically analyze the role of cell development models on tracking quality under increasing imaging intervals. Our results on a large 2D+t dataset demonstrate that model-driven cell tracking not only achieves higher accuracy at low frame rates, but also outperforms comparable methods in runtime efficiency. is available at https://github.com/JuBiotech/PyUAT , including example notebooks for immediate use in Google Colab.

Endoscopic swab sampling as a novel method for gastric Microbiome profiling: A pilot study compared with tissue biopsy

M-polynomial driven machine learning models for predicting physicochemical properties of antibiotics

Accurate prediction of the physicochemical properties of drug compounds is critical for the development of effective and safe antibiotics. In this study, we employ advanced machine learning techniques to address this challenge, using input data that includes M-Polynomials and various physicochemical descriptors. Three models were implemented: basic Support Vector Regression (SVR-Basic), optimized SVR (SVR-Tuned), and Random Forest (RF), trained on known compounds and tested on previously unseen drug samples to evaluate generalization. Model performance was comprehensively assessed using R 2 , MSE, RMSE, and MAE, alongside detailed error and residual analyses to ensure precision and robustness. Furthermore, residual-based metrics such as the Mean Residual (MR), Standard Deviation of Residuals (Std Residual), and Interquartile Range (IQR) of Residuals were employed to provide complementary insights into prediction bias, consistency, and robustness. By integrating feature importance analysis and ablation studies, the contribution of each molecular descriptor was systematically evaluated, providing deep insights into model stability and the key factors affecting predictive accuracy. Visual comparisons further illustrated the models’ behavior on training and test datasets. The results demonstrate that the proposed approach not only improves predictive accuracy compared to prior studies but also offers a robust and reliable framework for real-world drug development. All models were implemented in Python 3.12.7, highlighting the practical applicability of machine learning in pharmaceutical research.

Saving the locals: a conservation genomics approach to the endangered Spanish Toothcarp, Aphanius iberus (Valenciennes, 1846)

Abstract Understanding the genetic structure and evolutionary history of endangered species is crucial for effective conservation planning. The Spanish toothcarp, Aphanius iberus (Valenciennes, 1846), an endemic and euryhaline fish of the Mediterranean coast of the Iberian Peninsula, is currently threatened by habitat destruction, climate change, and anthropogenic translocations. Here, we employed both a single genetic marker (cytochrome b) and genome-wide SNP data from medium-coverage whole genomes to investigate the population structure, genetic diversity, and demographic history of A. iberus , especially focussing on its northern distribution, which has remained poorly studied. Our analyses revealed a well-structured genetic pattern across the species’ range, with four main genetic lineages: Northern Catalonia, Southern Catalonia, Levantine, and Murcian. Genomic indicators, including heterozygosity, ROHs, and migration analyses, suggest higher inbreeding and genetic erosion in the northernmost populations, likely due to long-term isolation, whereas southern populations maintain higher genetic diversity. We also identified several admixed and potentially translocated populations. These findings underscore the importance of accurately determining the origin of populations before any translocation or reintroduction, as misguided management may compromise the genetic integrity of native lineages. This work provides essential genomic insights to guide conservation strategies and emphasizes the need for lineage-aware management of endemic species like A. iberus .

Do ‘leaders’ in change sound different from ‘laggers’? The perceptual similarity of New Zealand English voices

Work on covariation in New Zealand English has revealed groups of speakers characterised by their back vowel spaces and status as ‘leaders’ or ‘laggers’ across a set of ongoing vowel changes. We investigate whether listeners hear speakers from different groups as perceptually distinct. We conduct a perception task in which New Zealanders rate the similarity of pairs of speakers. We use the results to create a two-dimensional perceptual similarity space by means of Multi-Dimensional Scaling, and test if speakers are organised within this space according to their back vowels, leader-lagger status, speed, or mean pitch. Results indicate higher pitched and faster speakers are perceptually distinct from lower pitched and slower speakers. Leaders are perceptually distinct from laggers if they are not markedly higher pitched. A Generalised Additive Mixed Model fit to the trial-by-trial ratings shows order effects, revealing that perception of similarity is not symmetrical. They also support the perceptual relevance of speaker speed, pitch and leader-lagger status.

A novel deep learning based spatial ensemble approach and segment anything model for landslide risk assessment in Chamoli district of Garhwal Himalayas

Acute kidney injury and chronic kidney disease in Chile: Temporal trends in hospitalization rates from 2010 to 2019

Background Kidney diseases pose considerable challenges to health systems affecting millions of people worldwide. In Latin America there are limited epidemiological studies on these conditions. Objective This study aimed to describe the national age-adjusted hospitalization trends for acute kidney injury (AKI) and chronic kidney disease (CKD) in the Chilean population. Methods We conducted a retrospective study using a secondary data from hospitalized cases diagnosed with AKI and CKD in Chile, from 2010 to 2019. Age-adjusted hospitalization rates and JoinPoint regression analysis were performed. Results Between 2010 and 2019, 26,715 hospitalizations of patients diagnosed with AKI and 99,816 patients with CKD were recorded. Nationally, long-term trends in age-adjusted hospitalization rates for AKI increased by 9.1% annually (95% CI: 8.0 to 10.5), whereas CKD hospitalization rates declined by 4.35% (95% CI: −5.76 to −2.99) over the study period. Short-term trends in AKI hospitalization rates showed initial declines in Maule between 2010–2017 (−7.97%, 95% CI: −13.63 to −6.04) followed by increases from 2017 to 2019 (8.45%, 95% CI: −4.08 to 17.66), whereas Valparaíso (7.2%, 95% CI: 5.2 to 9.5) experienced steady growth. Although CKD rates generally declined, central regions such as O’Higgins and Maule showed increases between 2016–2019 (10.6%, 95% CI: 5.1 to 19.9) and 2017–2019 (8.5%, 95% CI: −4.1 to 17.7), respectively. Conclusion/Interpretation This study highlights AKI as a growing public health challenge nationwide, with increasing hospitalization rates. At the same time, CKD hospitalizations show a reduction, although regional disparities remain a concern. Efforts should focus on enhanced integral surveillance to address these trends effectively across the country.

Designing a residual-enhanced hybrid Prophet–LSTM framework for urban air pollution forecasting in Beijing

Delayed delivery of antibiotics by ultrasound-mediated rupture of polylactic acid pockets: In vitro and in vivo studies

Surgical site infections are a devastating complication of instrumented orthopaedic surgery, particularly in the spine. Bacterial biofilms, once formed on the implant surfaces, exhibit antibiotic tolerance and immune escape, which lead to treatment challenges. Up to 9% of instrumented spine surgeries result in infection, despite the use of systemic antibiotics and local, powdered antibiotics for prophylaxis. The bacteria that survive the initial prophylaxis may be susceptible to a second, high dose of antibiotic prophylaxis prior to establishing at the site. Hence, we designed a local drug delivery system consisting of polylactic acid (PLA) film pockets that can be noninvasively triggered ( i.e. , ruptured) by the external application of ultrasound (US) following a delay of up to 6 days. We found that thin PLA films (24 ± 4.5 µm) with embedded vancomycin (VAN) powder assembled in a conical pocket shape best allowed for stability and rupturability of this US-mediated delivery system in vitro (92% US-triggered rupture of PLA-VAN pockets vs 31% of neat PLA pockets, p  < 0.0001). VAN-embedded PLA films exhibited decreased strength and toughness (3.47 J/mm 3 vs 0.68 J/mm 3 , p  < 0.0001) and surface VAN was rapidly released upon submersion, adding an additional layer of protection against bacterial colonization of the device. Finally, a pilot in vivo study in five rabbits demonstrated the feasibility of the design, but the stability of neat PLA and PLA-VAN pockets were varied (9/15 of all pockets were intact by Day 3). All of the 5 intact pockets (3 neat PLA and 2 PLA-VAN) that were allocated for insonation ruptured following US (100%). Overall, this design has the potential for use in targeting orthopaedic infections as well as for US-triggered bolus drug release for prophylaxis in high risk cases.

A multidimensional analysis of the 21st century competencies scale through ai-driven data mining techniques

Abstract In recent years, evaluating competencies such as knowledge, practical skills, character traits, and meta-learning capabilities has gained increasing importance in educational research. As educational datasets grow larger and more complex, machine learning offers promising tools for analyzing student responses and identifying patterns that support assessment processes. This study aims to classify student responses collected through the 21st Century Competencies Scale using a variety of machine learning algorithms, including SVM, ANN, k-NN, RF, LR, DT, AdaBoost, Gradient Boosting, and XGBoost. The dataset contains responses from 616 participants and covers four key sub-dimensions. Model performance was measured using accuracy, precision, recall, and F1-score. Grid search optimization was also applied to improve performance. The highest classification accuracy was achieved by LR in the “Character” sub-dimension (78.73%), followed by SVM in the “Skills” (78.58%) and overall scale (74.51%). Gradient Boosting and k-NN models also showed competitive results across multiple dimensions. These findings emphasize the effectiveness of machine learning, particularly when combined with parameter optimization, in supporting data-driven educational assessments.

Global trends and future predictions of pancreatic cancer incidence and mortality from 1990 to 2030: A comparative analysis of China, Korea, Japan, and the USA

Background Pancreatic cancer is a leading cause of cancer-related deaths globally, with rapidly increasing incidence and mortality rates in several regions. This study provides a comparative analysis of pancreatic cancer trends in China, Korea, Japan, and the USA, and projects these trends through 2030. Methods Data from the Global Burden of Disease (GBD) study were used to examine trends in age-standardized incidence rates (ASIR), age-standardized mortality rates (ASMR), and disability-adjusted life years (DALYs) from 1990 to 2020. Joinpoint regression and age-period cohort (APC) models were employed to assess temporal trends. Future predictions of ASIR, ASMR, and DALYs were generated using predictive modeling to extend the analysis to 2030. Results The study found that Japan and the USA have the highest incidence and mortality rates for pancreatic cancer, with predicted ASIR values expected to exceed 15 per 100,000 in Japan by 2030. Mortality trends follow a similar pattern, with males consistently showing higher ASMR than females across all regions. China is expected to see continued growth in DALYs, particularly among males, while Korea exhibits stable or slightly declining trends. The longitudinal age curves indicate that pancreatic cancer rates rise significantly with age, especially in individuals over 65 years. By 2030, Japan and the USA are projected to experience the greatest increase in pancreatic cancer burden, while Korea’s trends remain relatively stable. Conclusions Pancreatic cancer continues to pose a significant public health challenge, particularly in aging populations. The predicted rise in incidence, mortality, and DALYs in Japan and the USA calls for urgent implementation of targeted screening and prevention strategies, particularly for older age groups and high-risk populations.

Generalizable CNN-based detection of form and speed errors in physiotherapy using multimodal activity images

The effects of 12 weeks of chiropractic spinal adjustments on physiological biomarkers in adults: A pragmatic randomized controlled trial

Background Longer-term effects of chiropractic care on neuroplasticity, stress, and immune biomarkers remain unclear. Objective This study evaluates the effects of chiropractic care on physiological biomarkers, including brain-derived neurotrophic factor (BDNF), cortisol (saliva, blood, hair), and inflammatory cytokines [interleukin-6 (IL-6), tumor necrosis factor-alpha (TNF-α), interferon-gamma (IFN-γ), C-reactive protein (CRP), B-lymphocytes (CD19), T-helper cells (CD4), cytotoxic T cells (CD8), and natural killer cells (CD56)] in subclinical spinal pain patients. Methods Parallel-group, pragmatic randomized controlled trial conducted at the Rehabilitation Center of Railway General Hospital, Rawalpindi, Pakistan. Intervention: 12 weeks; follow-up: 16 weeks (May–December 2022). Participants with subclinical spinal pain were randomly assigned by using simple lottery method to either 12 weeks of chiropractic or sham care. We aimed to recruit up to 150 participants over three months; however, given the pragmatic nature of the trial and logistical constraints, including the availability of chiropractors, the final number enrolled was determined by how many eligible participants could be recruited during this time. Adults aged 20–60 years with subclinical spinal pain (n = 106 randomized; 88 completed 12-week measures; 73 completed 16-week follow-up). Among those who finished 12 weeks: chiropractic, 26 males/15 females, mean age 37.49 ± 12.39 years; sham, 24 males/23 females, mean age 26.85 ± 7.13 years. The primary outcome blood BDNF and secondary outcome, including saliva, blood and hair cortisol, IL-6, TNF-α, IFN-γ, CRP, CD19, CD4, CD8, and CD56 levels were measured at baseline, after 12 weeks of intervention, and at a 16-week follow-up. Linear and linear mixed-effects regression models were used to assess the effect of care and time on biological measures. Results Significant between-group differences were observed after 12 weeks of intervention, with higher salivary cortisol 5 ± 2 [0, 10], p = 0.045 and blood BDNF150 ± 60 (40, 270), p = 0.009 and IL-6 1.0 ± 0.3 [0.5, 1.5], p < 0.001 levels in the chiropractic care group. At the 16-week follow-up, blood cortisol −9 ± 4 [−17, −1], p = 0.024, IFN-γ − 22 ± 7 [−35, −9], and TNF-α −2 ± 1 [−5, 0], p = 0.028 levels increased in the sham group. Within-group comparisons showed a non-significant 10 ± 20 [−20, 50], p = 0.439 reduction in hair cortisol levels in the chiropractic group at 12 weeks, along with increased levels of blood cortisol, BDNF, CD8, CD4, IL-6, and CD19. Conclusion 12 weeks of Chiropractic care modulates biomarkers linked to neuroplasticity, inflammation, and stress. Increases in brain-derived neurotrophic factor and interleukin-6 suggest enhanced neuroplasticity and inflammatory responses, while decreases in tumor necrosis factor-alpha indicate a regulatory effect on systemic inflammation. These findings support the notion that chiropractic care modulates physiological systemic biomarkers, which may underscore its benefits on clinical outcomes. Trial registration ClinicalTrials.gov NCT05369156 .

Extending Schelling’s segregation model from self orientation to social orientation

Abstract Schelling’s segregation model demonstrates how simple local rules can generate large-scale social patterns, yet it assumes agents act myopically and ignore the broader consequences of their moves. We extend this framework by introducing social orientation as a behavioral microfoundation, reflecting humans’ communal nature. Socially oriented agents follow boundedly rational heuristics that account for the satisfaction of prospective neighbors within a two-step horizon of observability, in addition to their own. We formalize this through three relocation rules—negative externality avoiding (NEA), positive externality favoring (PEF), and positive externality optimizing (PEO)—each capturing a different balance between minimizing disruption and promoting stability. Agent-based simulations reveal that these rules, while maintaining satisfaction, consistently reduce segregation, accelerate the attainment of global stability, and lower relocation moves per agent, thereby reducing social costs. These results provide a stronger behavioral foundation for segregation modeling and show how locally rational, socially sensitive decision-making can scale into equilibria that are not just welfare-enhancing but also yield more integrated and resilient communities.

Expression of Concern: Ginsenoside Metabolite Compound K Promotes Recovery of Dextran Sulfate Sodium-Induced Colitis and Inhibits Inflammatory Responses by Suppressing NF-κB Activation

AI chatbots can persuade voters to change their minds

Brain network and activity in frontal regions change in the process of visuomotor adaptation

Abstract Frontal regions are involved in early stages of motor adaptation. However, the interactions between prefrontal cortex (PFC) and motor cortex (MC) remain unclear. Therefore, we investigated the relationship between PFC and MC in the process of motor adaptation including the early stage, using analyses of brain activity and its effective connectivity (EC). To promote motor adaptation, right-handed participants were trained to perform a visuomotor task with their right hand. To investigate the involved brain networks, their brain activity before and after training was measured by near-infrared spectroscopy (NIRS). EC analysis revealed that the brain network connecting the right PFC and contralateral (left) MC attenuated after the visuomotor training. Brain activity analysis showed the significant decrease in the ipsilateral (right) primary motor cortex (M1) after the visuomotor training. These results suggest that the PFC and MC are synchronized at the early stages of motor learning but act independently through the motor adaptation.

A compact design of four-port high-gain MIMO antenna using hybrid coupler and dual-polarized radiators

This paper introduces a design methodology for a multiple-input multiple-output (MIMO) antenna that achieves multi-port operation, high gain, as well as compact dimensions. The proposed design employs two dual-polarized radiators integrated with two hybrid couplers. This architecture enables a four-port MIMO array to be realized using only two radiating elements. Meanwhile, the hybrid couplers simultaneously excite both ports of the dual-polarized elements with equal magnitude and a 90 ∘ phase difference, facilitating high-gain radiation. To validate the proposed concept, an antenna prototype was fabricated and tested. Measurements confirm that the antenna, with an overall compact size of 0.96 λ × 0.77 λ × 0.04 λ , achieves a 3.1% operating bandwidth (4.72-4.87 GHz) with inter-port isolation better than 10 dB. Within this band, the antenna maintains a measured gain of around 8.0 dBi. Additionally, the antenna also performs good diversity performance in terms of Envelop Correlation Coefficient, Diversity Gain, Channel Capacity Loss, and Mean Effective Gain.